While the majority of Genscape’s proprietary network consists of electromagnetic field (EMF) monitors, they are not the only way that Genscape is able to capture critical market data. When EMF monitors cannot be deployed, Genscape turns to install infrared (IR) cameras in the field to help gather insightful intelligence for our customers across the globe. Genscape’s IR cameras are often deployed when the transmission lines are underground, the surrounding area is mostly industrial, or the transmission lines are in a crowded right of way.
While valuable, Genscape IR data has historically been less accurate than our EMF data. Due to this discrepancy, Genscape power market analysts have taken significant steps to improve the way customers interact with the IR data as well as the overall accuracy in order to enhance the image’s appeal to the human eye. One of the first IR process improvements was to streamline the way images are visually presented at five-minute intervals. Users can now easily track and playback any event on the cameras of interest by leveraging the new IR Dashboard on the PowerRT platform. Access to the image allows users to verify an event seen in the data or to determine if there are factors causing the data to skew.
In order to minimize the time that customers may have taken to frequently check individual images, Genscape analysts sought to reduce the amount of noise that the legacy calibration process produced against what was apparent to the human eye.
Our internal team used the latest in machine learning to build a new tool to make improvements to the calibration process. The main focus of this effort was to reduce the amount of manual intervention needed in calibrating an IR camera. Before these changes, if an IR monitored plant needed to be re-calibrated it could take a week for even the simplest update because of the manual work involved. Now with the improved calibration algorithm, a simple re-calibration can be done in less than one day and can sometimes be completed in a few hours. This effort not only helps Genscape analysts process a lot of data internally, but also gives customers better data at faster rate.
Using this new calibration algorithm, our analysts can now separate images into on and off datasets within the tool. Thousands of images can be used in the training to ensure the highest quality outcomes are achieved. The tool evaluates the majority of the image datasets for sources of truth for its defined state, including the heat signature and plumes produced by the monitored unit. Once the evaluation process is complete, the program takes the remaining images in the dataset to test itself. This process is repeated multiple times and as the algorithm goes through each learning iteration it improves to determine the images that represent both on and off states.
Through this new tool, we were able to significantly improve how we report on weather, which is notorious for skewing IR data. Since the human eye is helping to train the tool to identify an on or off signal in various states, the algorithm has been “taught” how to handle weather in this same way.
The Genscape power market analysts have also researched how these units typically run as part of this calibration overhaul. In doing so, they are able to accurately capture the level of output at each plant based on historical output as it relates to corresponding images. Due to these improvements to the calibration process and algorithm, the accuracy stems from fewer megawatt fluctuations and clearer on or off designations.
Genscape has already started to calibrate new power plants covered by IR cameras with this process and is in the midst of converting older cameras to support these new calibration methods. We are committed to improving the precision of our data so that our customers can benefit in their daily analysis of market opportunities. To learn more, or to request a trial of Genscape’s PowerRT platform, please click here.